This is a follow-up to #12640.
While profiling latency of working with a project with 8192 diagnostics
I've noticed that while we're parsing the LSP messages into a generic
message struct on a background thread, we can still block the main
thread as the conversion between that generic message struct and the
actual LSP message (for use by callback) is still happening on the main
thread.
This PR significantly constrains what a message callback can use, so
that it can be executed on any thread; we also send off message
conversion to the background thread. In practice new callback
constraints were already satisfied by all call sites, so no code outside
of the lsp crate had to be adjusted.
This has improved throughput of my 8192-benchmark from 40s to send out
all diagnostics after saving to ~20s. Now main thread is spending most
of the time updating our diagnostics sets, which can probably be
improved too.
Closes #ISSUE
Release Notes:
- Improved app responsiveness with huge # of diagnostics.
Follow-up of https://github.com/zed-industries/zed/pull/23112
Same reasoning applies.
Release Notes:
- Changed default formatter for C to be the primary language server, not
Prettier. Format-on-save is still disabled by default for C, but if one
uses the editor: format command now, it will default to the language
server. clangd can format C files, whereas prettier cannot.
As @hferreiro points out in [this
comment](https://github.com/zed-industries/zed/pull/18752#issuecomment-2589340565):
C++ and prettier don't work well together, so let's make the default
formatter for C++ the primary language server. We get that by disabling
prettier.
Release Notes:
- Changed default formatter for C++ to be the primary language server,
not Prettier. Format-on-save is still disabled by default for C++, but
if one uses the `editor: format` command now, it will default to the
language server. `clangd` can format C++ files, whereas prettier cannot.
This PR is a follow-up to
https://github.com/zed-industries/zed/pull/22911 to further improve the
registration of code action providers for the Assistant in order to
prevent duplicates.
The `CodeActionProvider` trait now has an `id` method that is used to
return a unique ID for a code action provider. We use this to prevent
registering duplicates of the same provider.
The registration of the code action providers for Assistant1 and
Assistant2 have also been reworked. Previously we were not call the
registration function—and thus setting up the subscriptions—until we
resolved the feature flags. However, this could lead to the registration
happening too late for existing workspace items.
We now perform the registration right away and then remove the undesired
code action providers once the feature flags have been resolved.
Release Notes:
- N/A
This PR fixes an issue with jumping from multi_buffer to a file; namely,
the scroll offset of the opened buffer used to match the position within
the multibuffer, but it broke a while back. This is because we were
opening a buffer without providing the data about the origin scroll
offset.
Closes #ISSUE
Release Notes:
- Fixed a bug where the relative position of an excerpt within the
multibuffer was not accounted for while jumping to the buffer, causing
the clicked line to drastically change position on screen.
The new `ContextMenu`-based `ContextPicker` requires initialization when
opened, but we were only doing this for the `ContextStrip` picker, not
the inline one.
Additionally, because we have a wrapper element around ContextMenu, we
need to propagate the `DismissEvent` so that it properly closes when
Escape is pressed.
Release Notes:
- N/A
These comments are inaccurate. Even if `convert_case` provided a way to
customize which boundaries were used (which is now does, it 0.7.1), they
would be removed from the string and replaced with the new boundary
character (`-`, `_`, ...), and we'd lose the ability to reconstruct the
text the way the author formatted it. This is not a hack, this is the
way we have to do it.
Release Notes:
- N/A
In current code this doesn't have benefit. In preparation for avoiding a
clone of workspace configuration. Having the interface this way may make
opportunities for efficiency clearer in the future
Release Notes:
- N/A
rust-analyzer does not support derive_macro expansion in attributes -
https://github.com/rust-lang/rust-analyzer/issues/8092. This could be
worked around via a proc_macro, but I think it'd be best to just require
docs for every action.
Release Notes:
- N/A
TODO:
- [ ] Investigate incorrect hit target for `stage all` button
- [ ] Add top level context menu
- [ ] Add entry context menus
- [x] Show paths in list view
- [ ] For now, `enter` can just open the file
- [ ] 🐞: Hover deadzone in list caused by scrollbar
- [x] 🐞: Incorrect status/nothing shown when multiple worktrees are
added
---
This PR continues work on the feature flagged git panel.
Changes:
- Defines and wires up git panel actions & keybindings
- Re-scopes some actions from `git_ui` -> `git`.
- General git actions (StageAll, CommitChanges, ...) are scoped to
`git`.
- Git panel specific actions (Close, FocusCommitEditor, ...) are scoped
to `git_panel.
- Staging actions & UI are now connected to git!
- Unify more reusable git status into the GitState global over being
tied to the panel directly.
- Uses the new git status codepaths instead of filtering all workspace
entries
Release Notes:
- N/A
---------
Co-authored-by: Cole Miller <53574922+cole-miller@users.noreply.github.com>
Co-authored-by: Cole Miller <cole@zed.dev>
Closes#23006
This PR should have been split into two, but since the changes are
related, I merged them into one.
1. On load, the title bar actions and bottom bar toggles are
unresponsive until the center pane is clicked. This happens because the
terminal captures focus (even if it's closed) long after the workspace
sets focus to itself during loading.
The issue was in the `focus_view` call used in the `new` method of
`TerminalPanel`. Since new terminal views can be created behind the
scenes (i.e., without the terminal being visible to the user), we
shouldn't handle focus for the terminal in this case. Removing
`focus_view` from the `new` method has no impact on the existing
terminal focusing logic. I've tested scenarios such as creating new
terminals, splitting terminals, zooming, etc., and everything works as
expected.
2. Currently, on load, docked terminals do not automatically focus when
they are only visible item to the user. This PR implements it.
Before/After:
1. When only the dock terminal is visible on load. Terminal is focused.
<img
src="https://github.com/user-attachments/assets/af8848aa-ccb5-4a3b-b2c6-486e8d588f09"
alt="image" height="280px" />
<img
src="https://github.com/user-attachments/assets/8f76ca2e-de29-4cc0-979b-749b50a00bbd"
alt="image" height="280px" />
2. When other items are visible along with the dock terminal on load.
Editor is focused.
<img
src="https://github.com/user-attachments/assets/d3248272-a75d-4763-9e99-defb8a369b68"
alt="image" height="280px" />
<img
src="https://github.com/user-attachments/assets/fba5184e-1ab2-406c-9669-b141aaf1c32f"
alt="image" height="280px" />
3. Multiple tabs along with split panes. Last terminal is focused.
<img
src="https://github.com/user-attachments/assets/7a10c3cf-8bb3-4b88-aacc-732b678bee19"
alt="image" height="270px" />
<img
src="https://github.com/user-attachments/assets/4d16e98f-9d7a-45f6-8701-d6652e411d3b"
alt="image" height="270px" />
Future:
When a docked terminal is in a zoomed state and Zed is loaded, we should
prioritize focusing on the terminal over the active item (e.g., an
editor) behind it. This hasn't been implemented in this PR because the
zoomed state during the load function is stale. The correct state is
received later via the workspace. I'm still investigating where exactly
this should be handled, so this will be a separate PR.
cc: @SomeoneToIgnore
Release Notes:
- Fixed unresponsive buttons on load until the center pane is clicked.
- Added auto-focus for the docked terminal on load when no other item is
focused.
These keybindings extend the already selected text. This allows closer
emacs emulation where subsequent movement commands extend / shrink the
current selection instead of dismissing it.
This is a follow up on
- #21927
Release Notes:
- Added emacs movement keybindings that extend/shrink the current
selection
---------
Co-authored-by: Peter Tripp <peter@zed.dev>
Also:
* Adds `impl_internal_actions!` for deriving the `Action` trait without
registering.
* Removes some deserializers that immediately fail in favor of
`#[serde(skip)]` on fields where they were used. This also omits them
from the schema.
Release Notes:
- Keymap settings file now has more JSON schema information to inform
`json-language-server` completions and info, particularly for actions
that take input.
I added these notifies in #23011, but in practive have found them to be
overly disruptive. It would definitely be good to do something better
than logging here, but having a sticky error notification is worse. I
think it is still good to notify on mutation failures, so left those in
In particular with rust-analyzer, "Go to definition" and "Find
references" frequently fail with "Content modified" quite a while after
sending the request. Since users are probably used to these operations
being finicky it doesn't seem useful to have a prominent display of
errors for them.
It seems the original author intended to write either "`ctrl+c` to copy"
or "`ctrl+v` to paste". Updated to be "`ctrl+v` to paste".
Release Notes:
- N/A
Co-authored-by: Michael Sloan <michael@zed.dev>
This adds support for LSPs that use the old rename flow which does not
first ask the LSP for the rename range and check that it is a valid
range to rename.
Closes#16663
Release Notes:
* Fixed rename symbols action when the language server does not have the
capability to prepare renames - such as `luau-lsp`.
Not sure why scroll was janky with `Autoscroll::newest()`, but this
appears to fix it. Probably better to conditionally do the autoscroll
requests anyway.
Release Notes:
- N/A
Closes#18737
Release notes:
- Improved LSP debug logs by defaulting to soft wrap and folding a
suffix of long lines. Also adds autoscroll, so if the cursor is on the
last line of the logs they will scroll like `tail`.
- Move function queries under constant queries to avoid uppercase
functions highlighted as constants
- Merge keywords and remove duplicates
- Highlights type aliases on import
- Highlights literal built-in types (null, undefined, true, false) as
`@type.builtin`
Confused about case-based queries, should they be rewritten?
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
This PR also removes the `ThumbsUp` action that wasn't being triggered
correctly. We didn't have it's counterpart `ThumbsDown`, too, so I
mostly assumed it would be harmless to remove `ThumbsUp` as well.
<img width="800" alt="Screenshot 2025-01-10 at 6 18 44 PM"
src="https://github.com/user-attachments/assets/9fd5da9f-9dff-454d-9f31-c02f1370b937"
/>
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
